Point Cloud Geometry Coding with LIDAR-Based Coordinate Revision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional point cloud coding techniques suffer from inefficiencies in geometry coding, particularly in sparse distributions, leading to distortion in x, y, and z coordinates, despite having potential information from LIDAR data that could reduce this distortion.
Innovation Solution
The proposed method involves determining the capturing laser for each point in the point cloud and using its elevation or azimuthal angle to revise the z or x/y coordinates, respectively, through functions like tan(x^2 + y^2) or tan(x^2 + y^2) + zs, where zs is the laser head position shift, to improve coordinate precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional point cloud coding techniques are used for geometry coding, then the coding process is simple, but coordinate distortion occurs and coding efficiency is poor
Solution Approach 1:
The patent applies de-quantization to coded geometry coordinates before attribute coding, which is a preliminary action that restores coordinate precision before subsequent processing. This preliminary restoration of precision improves both measurement accuracy and overall coding efficiency by preventing error propagation in sparse point cloud distributions
Solution Approach 2:
The patent changes the parameter state of geometry coordinates by applying de-quantization, transforming coded coordinates with distortion into restored coordinates with improved precision. This parameter change resolves the contradiction by improving coordinate precision while maintaining coding efficiency through the restored accuracy
2Measurement precision
If de-quantization is applied to coded geometry coordinates before attribute coding, then coordinate precision is improved, but processing complexity increases
Solution Approach 1:
De-quantization is performed as a preliminary action before attribute coding, restoring geometry coordinate precision early in the process. This approach improves measurement precision while managing complexity by addressing the distortion issue before it propagates through subsequent coding stages
Solution Approach 2:
The patent incorporates feedback by using restored geometry coordinates from de-quantization to improve subsequent attribute coding. This feedback loop ensures that precision improvements are maintained and propagated through the coding process, justifying the additional processing complexity
3Measurement precision
If LIDAR data information is utilized to revise coordinates, then coordinate distortion is reduced, but processing time increases
Solution Approach 1:
The patent applies coordinate revision using LIDAR data information as a preliminary action to reduce distortion before attribute coding. This preliminary correction improves coordinate accuracy while managing processing time by addressing distortion early rather than requiring extensive post-processing
4Productivity
If conventional coding is used without de-quantization, then processing is faster, but coding efficiency and accuracy deteriorate
Solution Approach 1:
De-quantization is applied as a preliminary action to restore geometry coordinate precision before attribute coding. This preliminary restoration improves coding efficiency and accuracy by preventing error propagation, demonstrating that the additional processing step enhances overall productivity rather than hindering it
Data Source
AI summary
Embodiments of the present disclosure provide a method for point cloud coding. In the method, for a conversion between a current coding unit of a point cloud sequence and a bitstream of the point cloud sequence, at least one coded geometry coordinate of the current coding unit is determined. A de-quantization is applied to the at least one coded geometry coordinate. An attribute coding is applied to the at least one de-quantized geometry coordinate of the current coding unit. The conversion is performed based on the attribute coding.